Hi Andreas,
Is the column following the DPA value the statistical error? I found that if the bin is set to larger value, this error becomes larger.
That is correct, the statistical error is expressed as percentage with respect to the value in the preceding column. Concerning the second statement: in general if you increase the bin size the error should go down since you are collecting more statistics per bin. However this goes at the cost of accuracy since averaging of the obtained values obscures the distribution inside one bin, in this case if the dpa is not uniformly distributed inside the bin volume. So smaller bins on the other hand would in theory give better resolution but can also lead to higher errors because of the small amount of statistics included, the profile of results can therefore show large fluctuations (when plotting using Flair for example).
1 Like
Hi Andreas,
I found that smaller bins (10 um size) can lead to error as high as 99%. Can these results be used? Is there any method to decrease the error?
As mentioned above this can happen because the amount of statistics obtained per bin is small. Here the most straightforward ways to remedy this is by either 1. running more primary particles to collect more statistics (if this is feasible from a computational time point of view) until the errors are reduced to an acceptable size, or 2. making the bin sizes larger again. When doing this you have to be careful to make sure that the values are still properly resolved, i.e. it can be assumed that inside one bin the dpa distribution is (close to) uniform so the final averaged result is physically representative.
1 Like